Steve Canepa’s Post

🚀 Exciting news: IBM Granite 3.0 is now running on POWER10! Following IBM's groundbreaking announcement of their latest AI models, we've already got these powerful models up and running on IBM POWER10 systems at TechXchange Las Vegas. The new Granite 3.0 family includes 8B and 2B parameter models, plus innovative MoE variants - all optimized for enterprise workloads and available under Apache 2.0 license. 🔍 Key highlights from the release: -> Up to 95% cost reduction per token generation -> State-of-the-art performance matching larger models -> Built-in safety mechanisms and RAG-specific capabilities -> Optimized for low-latency environments 📣 Join us TOMORROW at TechXchange 2024 in Las Vegas! Session 1599: "Deploy LLM AI Model on Power and use it for Inferencing" Time: 9:00 AM PDT Instructors: Daniel de Souza Casali, Jason Liu, Rodrigo Ceron, and myself We'll walk you through configuring LLM and RAG on IBM Power systems. Don't miss this opportunity to learn how to leverage these powerful models in your enterprise environment! The photo below was taken after Tuesday's hands-on lab. #IBMPOWER #AI #MachineLearning #EnterpriseAI #IBMTechXchange #GraniteAI #Innovation

Regardless of the domain, organizations striving to become more data-driven and incorporate AI and GenAI into their operations require a robust data platform. IBM Storage Scale, with its integrated Data Cataloging Service, offers key enablement services essential for modern businesses: - Data Discovery: Enable data discovery through enterprise search with deep indexing, allowing for efficient research and productivity. - Data Agility: Ensure seamless data access from various environments and high-performance data caching for optimal application performance. - Data & AI Performance: Maximize data throughput to GPU-accelerated frameworks, leveraging the right storage and networking solutions. - Data Resilience: Safeguard training data against cyber threats to maintain data integrity and model security. IBM Storage Scale not only provides core data services but also delivers record-breaking performance to fuel your AI Center of Excellence. It supports both IBM and non-IBM storage, ensuring high-performance data processing for new and existing datasets. Deployable on major cloud platforms like Azure, AWS, and GCP, IBM Storage Scale caters to cloud-native and on-premise workloads. Facilitating multi-site collaboration, it enhances agility, performance, and resiliency for data science and AI research teams worldwide. Learn more about IBM Storage Scale and its capabilities here: https://lnkd.in/gFs-DPQe

IBM 𝗦𝘁𝗼𝗿𝗮𝗴𝗲 𝗦𝗰𝗮𝗹𝗲 is software-defined file and object storage that enables organizations to build a global data platform for artificial intelligence (AI), high-performance computing (HPC), advanced analytics, and other demanding workloads.

Getting started quickly and easily with an AI cluster is important to many enterprise businesses today. While GPUs get a lot of the attention, the right networking design is also a critical part of a performant and cost effective solution. Join the Presidio and Intel teams for a one hour webinar on July 18th at 1-2pm EST as we cover a cluster concept, including networking and communication software, that is needed for a modest scale AI system that is both easily obtainable, and a logical extension to typical enterprise data center deployments. Topics: • Classes of AI systems and AI workloads • Ingredients of Enterprise AI Installation • Why a modest scale AI cluster is a great first step for most enterprises starting in AI  • Example cluster configuration, network and communication software stack https://lnkd.in/eFMppgbs #iamintel #presidio #ai #networking

Not just a new #SSD, but a BETTER SSD. Purpose built for the #AI era, the #CSD5000 delivers the performance, capacity, and efficiency needed for the AI/ML and data center infrastructure. Check it out https://buff.ly/46xPFCb #10YearAnniversary #storage #innovation

In 2011, IBM's AI, Watson, gained international attention after beating two champions on Jeopardy. In 2011 Watson was installed in 10 racks, powered by a cluster of 90 servers with a total of 2,880 processor cores, 15 terabytes of RAM, and 500 gigabytes of pre-processed data. It will be cool to get IBM to build the same system on modern hardware. It will most likely run in two racks, 20 Servers Power1050, 15TB RAM, F9200 with 2PB storage. " "Throwback to 2011 when IBM's AI, Watson, made headlines by defeating two Jeopardy champions. With 10 racks, a cluster of 90xP750 servers (2,880 processor cores), 15 terabytes of RAM, and 500 gigabytes of pre-processed data, Watson was a game-changer. Imagine the advancements we can achieve with IBM building the same system on modern hardware! With two racks, 20 Power1050 servers, 15TB RAM, and F9200 with 2PB storage, the potential for groundbreaking AI capabilities is mind-boggling. Exciting times ahead indeed!

Curious how watsonx ensures being open, trusted, targeted and empowering? 🧐 IBM watsonx Assistant‘s embedded LLMs are customized IBM Granite models. Give the Publicly available data curation and data sources behind the IBM Granite models a read.

Unlock the power of real-time observability with IBM #Instana. Monitor and optimize your #applicationperformance like never before. Drive efficiency and elevate your digital landscape with precision insights. Discover the future of observability today and know more about #IBM Instana: https://lnkd.in/dPBqFCjf #IntelligentSystems #IBM #Instana #Observability #DigitalTransformation 

IBM Storage Scale day in NY saw an enlightening session by Matt Klos, sharing valuable insights on AI journey across hundreds of customers. Key takeaways included: - Multi-Protocol Data Access: Supporting simultaneous access and GPU Direct, enabling global collaboration regardless of location or format. - Storage Acceleration: Automatic caching of storage systems to speed up data queries and enhance economics. - Storage Abstraction: Providing a consistent experience with a single global namespace to boost efficiency and security. - Storage Assurance: Ensuring data security, governance, and integrity for seamless business continuity. #IBM #AI #IBMStorageScale #IBMScale #IBMAI #IBMResearch Matthew Klos Eleanor A. Lewin Stacie Mason Amy Hirst Scott Baker Sam Werner Denis Kennelly #IBMNewYork #Watson #NVIDIA

With security, resilience, hardware optimization, cost reductions all top priorities, IBM has built a way for your organization to truly harness AI with Fusion 2.8. Read this blog to accelerate your organization's digital transformation journey. #IBMPartners #IBMAmplify